Interactive Text-to-Image Retrieval with Large Language Models: A Plug-and-Play Approach

TL;DR

This paper introduces PlugIR, a plug-and-play method leveraging large language models for interactive text-to-image retrieval, eliminating the need for fine-tuning and improving question generation and retrieval performance.

Contribution

The paper presents a novel plug-and-play approach that reformulates dialogue queries and generates non-redundant questions using LLMs, enabling effective interactive retrieval without fine-tuning.

Findings

PlugIR outperforms zero-shot and fine-tuned baselines on benchmarks.

Introduces the BRI metric for comprehensive evaluation.

Demonstrates flexible application of the methodology.

Abstract

In this paper, we primarily address the issue of dialogue-form context query within the interactive text-to-image retrieval task. Our methodology, PlugIR, actively utilizes the general instruction-following capability of LLMs in two ways. First, by reformulating the dialogue-form context, we eliminate the necessity of fine-tuning a retrieval model on existing visual dialogue data, thereby enabling the use of any arbitrary black-box model. Second, we construct the LLM questioner to generate non-redundant questions about the attributes of the target image, based on the information of retrieval candidate images in the current context. This approach mitigates the issues of noisiness and redundancy in the generated questions. Beyond our methodology, we propose a novel evaluation metric, Best log Rank Integral (BRI), for a comprehensive assessment of the interactive retrieval system. PlugIR demonstrates superior performance compared to both zero-shot and fine-tuned baselines in various benchmarks. Additionally, the two methodologies comprising PlugIR can be flexibly applied together or separately in various situations. Our codes are available at https://github.com/Saehyung-Lee/PlugIR.